The Data Centre What?

04 Nov, 2026

Last week the UK Government published a policy paper, called The Data Centre Covenant. It wasn't signed by any minister or attributed to any part of the government machinery, though it did appear under the banner of the Treasury.

Many activists have commented on the emptiness of the reassurances given, the implausibility of many claims, and the large number of get out clauses which favour the builders of AI data centres. Here I want to focus instead on that word 'covenant'. Who chose that word? And what were they thinking? It's almost as if they asked an LLM to come up with a good title for their document. I assume it is intended to be reassuring. It isn't. It is ethics-washing.

Historical Aside

My first thought was of the Scottish Covenanters of 1638 and I assume anyone with a rudimentary knowledge of 17th century history would also have their mind turned to those brave, independent-minded people resisting the power of Whitehall. The seventeenth century was the last time we had a revolution in this country, and in fact we had two. The National Covenant of 1638 was one of the early events in the struggle against the Stuart monarchy, which led to the civil war and the abolition of the monarchy in 1649. And the centre of power of the Stuart monarchy was the Palace of Whitehall, where so many Government ministries, including the Treasury, are still based.

Covenants, then, can be seen through the lens of British history as symbols of rebellion against the overreach of state authority into matters which should be determined locally and as a matter of conscience. It is almost as if the anonymous authors of this document are unwittingly inviting data centre activists to formulate a new National Covenant, rejecting the authority of the state in these matters.

Definitions

So much for the historical associations. If we set those aside, we can just concentrate on what the word 'covenant' actually means. Being of a boringly academic mindset, my first port of call is always the Oxford English Dictionary.

Setting aside the theological meaning and the historical usage that I've already discussed, there are just two non obsolete definitions. Let's take the first:

1.a. A mutual agreement between two or more parties to do or refrain from doing certain acts. Sometimes also: the undertaking or pledge of either party in such an agreement. Frequently in to make (or enter into) a covenant, to hold (or keep, break) covenant. Now chiefly in legal or theological contexts.

A mutual agreement? Clearly this policy paper is not a mutual agreement between the Government and the citizens of this country, since they were not consulted, let alone had the opportunity to consent. A cynic might think that this is a mutual agreement between the Government and the lobbyists for the tech companies who want to accelerate the building of data centres in the UK. If this is the intended meaning, then the use is deceptive, for either it implies that there's been an agreement with the people of this country, which there hasn't, or it refers to an agreement behind closed doors with the corporate beneficiaries of government policy.

The other non-obsolete definition in the OED is the use of the term in law:

Law. A formal agreement or promise of legal validity; esp. (in English Law) a promise or agreement executed as a deed.

Again, if this is the intended meaning, the use is deceptive, for this policy paper has no such legal status. At best, this document serves as guidance for civil servants making decisions about the building of data centres.

Turning to the obsolete uses in the implausible hope that some erudite civil servant was attempting to resurrect one of them, we do find something more relevant:

2. A promise made to oneself, a personal pledge or vow.

This meaning removes the implications of mutual agreement, legal status, or the need for some form of security or collateral.¹ Thus it aligns better with what appears to be happening in this policy paper. However, to reintroduce this meaning into common usage is a pretty dramatic move since the OED's last recorded use in this definition is by Chaucer in ~1385.

LLMs have no sensitivity to the temporal dynamics of language, and thus might have suggested the word with this meaning 'in mind'. Why not simply 'Promise'? Presumably the Wordsmiths responsible for accepting the suggestion thought that 'Covenant' sounded more formal and serious.

However, if it is intended to be a rather obscure synonym for 'promise', we still have a fundamental problem with the document.

Promising creates a normative relation or directed duty between the person or organisation that makes the promise and the person to whom the promise is made. Furthermore, on the ordinary understanding of this duty, the promisee has a special status to claim or demand that the promisor does as promised and to expect some sort of compensation or at least sincere apology should they fail so to do.

Conclusion

In which case, the fundamental failing of this document is that it does not specify who is making the promise and to whom. It thus avoids the most basic feature of the ordinary understanding of a promise, which is that the promisor can be held to account for to do what they've promised to do. By not saying who can hold whom to account should the promise be unfulfilled, the document is worthless even if the details of the promise were substantial.